January 2017 saw a sensational piece of news: the Chinese newspaper Southern Metropolis Daily had published a piece written by a robot, for the first time in history. Robot writers — it certainly sounds very impressive.
Previously, the media loved the topic of artificial intelligence, but has now come face to face with the concept in its own field. Suddenly, publications have emerged about how the profession of journalism will soon disappear, people will stop being interested in the news and all journalistic authority will be handed over to robots.
Readers were also impressed: when you hear the phrase “robot journalists”, a fantastical version of reality springs to mind, where androids sit hammering keyboards, or there are newspapers with pages that are not only set but also composed by a printing press. And in the not-too-distant future, we could have an android winning the Pulitzer Prize for the first time.
The reality, however, is much more prosaic. The Chinese piece was dedicated to “tourism fever” regarding the Spring Festival travel rush to Guangzhou. It was very straightforward, recounting statistics over a reporting period and ending with simple summary of information that had already been provided: a large number of tourists came to Guangzhou because of the festivals.
Following the news from China, reports came that the British news agency Press Association planned to use artificial intelligence for covering future events across the United Kingdom and Ireland. Admittedly, the scope of activity for robot journalists is limited to the news, sporting announcements (about upcoming events, team victories and losses etc.) and election coverage (robots will simply process electoral Committee data and report on the results). In short, robots will perform simple data processing and monitor feeds.
Bots in the newsroom
In all fairness, the future of media is already here. Journalism bots already exist and are enjoying a certain degree of popularity. In a detailed review carried out by technology journal Wired, analyses showed that bots were more popular there than official news agencies. Their citation index is occasionally higher, and readers prefer them to paper publications. They report on events more quickly and are more concise and simple.
However, work produced by bots can contain pitfalls. As far as they are concerned, “there’s no need” to spend time referencing a news source, so the accuracy and reliability of the information that they deliver could be in doubt. Cases relating to this issue have somewhat increased lately: reporting on the death of a famous figure who is alive and well; the “leaking” of governmental decisions that have not yet been confirmed.
However, efficiency is the most valuable thing of all, and the “big” publications often prefer bots to classical news agencies. For them, the main thing is that they can report the news quickly. But disproving, deleting or making amends to a news report always takes time.
The media is going through hard times, and the issue of getting robots involved in journalism is becoming increasingly urgent. At the very least, newspapers and journals are forced to compete in terms of accuracy and efficiency on blogs and social media sites and integrate their services with these sites. And who could forget about the never-ending saga of fake news, alternative facts and the other demons plaguing contemporary American journalism. Robots and bots are being viewed as a panacea for all ills in this situation: they are reliable, efficient, and command high levels of trust among readers.
Down with the daily grind
The news report from Guangzhou is a great gift for editors and journalists. Artificial intelligence could liberate them from the most mundane parts of the job: searching, verifying and translating the news. Previously, each publication needed to employ interns and students who were willing to complete these tasks, monitoring news feeds and translating and relaying texts.
In the near future, it’s expected that this army of unskilled labourers will reject this type of work. Any publication can programme a news robot to search for news reports on a set number of topics – and news feeds become easy to scan.
The same robot may deal with “fact checking”, another time-consuming part of journalism. An algorithm can find sources which confirm this or that report. For example, before reporting on the death of a Nobel literature laureate (as was the case recently), an algorithm can find several other reports on this tragic event, but not copy the text of a tweet by an unknown prankster.
The Myth of the Robot Writer
The emergence of news reports created by robots hasn’t delighted everyone — and there are plenty of sceptics. Influential American scientific journalist John C. Dvorak published a column in PC Magazine entitled “The Myth of the Robot Writer”. In it, he left practically no stone unturned in his examination of all the fantasy scenarios that exist regarding the bright future of the press. The robot, he argues, can be a worthy assistant to the author of a text, but will never replace them. Another example – Wikipedia – has not made the paper encyclopaedia obsolete, but has simply forced the market to change.
It’s clear that some journalistic forms are simply beyond the capabilities of robots — for now, at least. Newspaper columns, recipes and analytical papers all demand more than just the ability to gather information from their authors. At the very least, you need to be able to put things in order – something that robots can’t do just yet. In addition, a journalist is required to apply a certain level of reflection, and evaluate collected material. Not to mention the fact that these genres prefer a distinctive, clearly expressed, recognisable style. The more journalism resembles literature, the more authorial it becomes and the less likely it is to become the domain of the robots.
There have been some attempts however, and there are now projects attempting to imitate the characteristic narrative style of this or that writer. The Higher School of Economics in Moscow is developing a programme that replaces distinctive words from the lexicons of Russian writers Leskov, Dostoevsky and Griboyedov with synonyms closely related in meaning. For example, the programme re-wrote the most famous quatrain from Alexander Griboyedov’s “Woe from Wit”, creating a very different passage to the original.
Anti-plagiarism and text generation
Yet all of this is only for entertainment. Robots perform the dirty, but unavoidable, work, and do it fairly well. Their most popular type of application is in the detection of plagiarism, and such programmes are mainly used by PR specialists. Algorithms detect anything that has been “borrowed” in a text (one of the paragraphs of this article, according to the programme, is 90% original material: the phrase “Nobel Prize for Literature” was flagged as a “borrowing”).
A simple query in any search engine will result in a wealth of generated text – in truth, they mainly create “filler” for web designers. Multiple attempts to generate minimally comprehensible text for use in this article were completely unsuccessful. The best that contemporary technology can offer is a defined number of random letters.
So the experts can continue to work in peace, as robots aren’t exactly going to steal their jobs any time soon. It is, however, time to start considering artificial intelligence as a colleague or a valuable assistant, as neither the scientific nor news industry could do without it.